Work package sizing and project performance

We study how design decisions in project planning affect the cost of execution. In organizing a project’s tasks into work packages, trade-offs arise. Defining small work packages increases project complexity and workload, and reduces economies of scale, whereas defining large work packages reduces concurrent processing and adversely affects cash flow. Our work is apparently the first to study this trade-off. We consider the objective of minimizing total project cost, subject to a deadline on project makespan. For serial task networks, we describe an efficient algorithm that finds optimal work package sizes. For acyclic task networks, we develop a heuristic method and a lower bound for the unary NP-hard problem. A computational study shows that our heuristic routinely delivers near-optimal solutions that substantially improve on those found by benchmark procedures. Our results demonstrate the value of deliberately varying work package sizes within a project, in contrast to typical project management practice. Related issues including multiple serial paths in parallel, task incompatibility, and generalized precedence constrained work packages are also discussed. Our work enables more precise planning of work packages to improve performance, documents the value of integrating the planning of work packages and schedules, and provides insights that guide resource allocation decisions.